IBM and Microsoft announced Imagine Enterprise on 10 November 2026. It is a platform for building, running and governing AI agents in regulated industries.
My view: use it as a benchmark, not a shopping list.
The real news is not the 6,000 agents. It is that two of the biggest names in enterprise technology are now selling governance as the product. That tells you where the market is going.
What IBM and Microsoft announced
The split is simple. IBM brings the agents and the industry knowledge. Microsoft brings the technology and the controls.
- IBM Consulting supplies more than 6,000 pre-built AI agents and assets, plus its industry expertise.
- Microsoft supplies the technology: Foundry, Copilot Studio, Fabric and Azure.
- The controls come from Microsoft Entra (who has access), Purview (data governance) and Defender (security).
- First industries: financial services, insurance and retail. Manufacturing, energy, public sector and mobility are planned.
- Engineers from both companies will work alongside client teams, but only on “qualifying engagements”.
- Three ways to buy: IBM runs it for you, you build it with IBM, or you manage it yourself.
Source: IBM and Microsoft joint announcement, 10 November 2026.
What is an AI agent, in plain English?
A chatbot answers a question. An agent does a job.
An agent can open a system, check a record, take the next step and pass the result to a person for approval. Put several agents together on one process and you get what vendors now call a “digital workforce”.
That is why governance matters more here than with a chatbot. A wrong answer is embarrassing. A wrong action can move money, reject a customer or break a rule.
How Imagine Enterprise fits together
There are three layers.
- The process layer. IBM’s tools, Process Studio, Context Studio and Advantage Core, connect agents to your own processes and data.
- The agent layer. Microsoft Foundry and Copilot Studio build and coordinate the agents.
- The control layer. Entra, Purview and Defender decide what each agent can access, track the data it touches and watch for threats.
The pitch is that you buy all three together, with industry agents already built, instead of stitching them together yourself.
What the announcement does not say
Read the release closely and four gaps stand out.
- No price. Nothing on cost or commercial model.
- No customer in production. SEI, the one named client, is described as exploring agentic technology with IBM and Microsoft. That is not the same as running it live.
- No definition of “qualifying”. We do not know who gets the engineers.
- Little detail on audit. The release promises controls and observability. It does not show how each agent action is logged for an auditor.
One more point, and this is my view, not the release. No platform takes regulatory accountability off your desk. If an agent gets a claim or a credit decision wrong, your firm answers for it.
Why leaders should care
This is not another model release. It is an attempt to make AI agents safe enough to run inside a bank, an insurer or a retailer.
1. Governance is now the selling point.
Boards are already asking who is accountable when an agent acts across systems. IBM and Microsoft are answering that question in the sales pitch. Expect every vendor to follow.
2. It could shorten the road out of pilots.
Pre-built agents and on-site engineers mean less building from scratch. That matters if you are short of AI talent. It is a promise for now, not a proven result.
3. The trade-off is dependence.
You move faster, but you tie yourself closer to IBM Consulting and the Microsoft stack. Many Irish firms already run on Microsoft 365, so the pull will be strong. Ask what it costs to leave before you ask what it costs to join.
4. Watch which euro you are chasing.
Agents on rules-heavy work, such as claims or onboarding, mostly deliver the Blue Euro: efficiency savings. That is worth having. The Green Euro, new revenue, still needs your own people and your own data.
The firms that win will treat agents as managed digital coworkers with clear limits. The ones that lose will either let Shadow AI spread or stay stuck in pilots.
What it means for Irish businesses
The first three industries, financial services, insurance and retail, are all big employers in Ireland. So this is relevant here from day one.
Two things make it sharper for Irish leaders.
- Regulation. The EU AI Act treats uses such as credit scoring and some insurance pricing as high risk. If your agents touch those decisions, you need the audit trail whether or not a vendor supplies it.
- Scale. This is built for large enterprises with IBM-sized budgets. A mid-sized Irish firm is unlikely to buy it. You can still copy the model: pre-agreed approval gates, full logging and one named owner for each agent.
Irish companies are already building governed agents
You do not have to wait for IBM and Microsoft to see this working. The four finalists for Agentic AI Innovation of the Year at the AI Awards 2026 are doing it now.
- Future Proof Knowledge runs agents on regulated care work, such as clinical deadlines, caregiver training and billing. Every task stays traceable.
- Oac Snacks built a multi-agent planning system for commercial, supply and finance decisions. Agents debate the evidence. A person approves the outcome.
- SPRYT has an agent called Asa that books, reschedules and reminds patients over everyday messaging channels. It cuts admin for healthcare teams.
- yieldHUB uses agents to monitor semiconductor test data and support decisions on the production line. The aim is higher yield and less retesting.
Look at the pattern. Clear agent roles. People kept in the loop. A record of what happened and why. That is the same model Imagine Enterprise is selling, built by Irish teams at a fraction of the scale.
Are you ready for agents? A 5Ps check
Before you talk to any vendor, score yourself on five questions. I use these with every leadership team.
- People. Who owns each agent, and who signs off its work?
- Process. Is the process written down well enough for an agent to follow it?
- Platforms. Do your systems let an agent act safely, with the right access and no more?
- Proprietary Data. Is the data the agent needs accurate, permitted and in one place?
- Products and Services. Where does this change what the customer gets, not just what it costs you?
If you cannot answer the first two, no platform will save you. Agents speed up a good process. They also speed up a bad one.
The Monday morning action plan
You do not need to buy anything to act on this. Three steps.
- Ask for the inventory.
Ask your CIO or head of AI for one page listing every agent and automation experiment in the business. Include the unsanctioned ones. Give them one week. - Run a one-week test on paper.
Pick one high-volume, rules-heavy process. Claims triage, vendor onboarding or policy exceptions all work. Write down the human approval gates, what must be logged, the success measures and who owns it when it goes wrong. That page is now your benchmark for every vendor. - Put five questions to any agent vendor.
- What does it cost, and how is it priced?
- Who in our sector runs it in production today?
- Can an auditor see every action an agent took, and why?
- Who is accountable when an agent gets it wrong?
- What happens to our agents and data if we leave?
If a vendor cannot answer all five, you are still in pilot territory.
Common questions
What is Imagine Enterprise?
A joint IBM and Microsoft platform for building, running and governing AI agents on industry processes. It was announced on 10 November 2026.
Who is it for?
Large organisations in regulated industries. Financial services, insurance and retail come first.
How much does it cost?
The announcement gives no price.
Is anyone using it yet?
No customer is named as running it in production. SEI is described as exploring it.
Should we wait?
Do not wait to get ready. The inventory, the one-week test and the 5Ps check cost nothing and apply to any platform you choose later.
Work through this with your leadership team
I help boards and leadership teams turn AI agents from a slide into a plan. If you want a keynote or a hands-on workshop on agents and governance, get in touch at markkellyai.com.
Share This Story, Choose Your Platform!
Planning an event or leadership offsite?
Mark Kelly delivers evidence-based AI keynotes and hands-on AI leadership workshops for boards, executive teams and conferences across Ireland and Europe.





